‘I did it because…’: pedestrian decision-making at road crossings in a longitudinal Virtual Reality study with automated vehicles

Vulnerable Road Users (VRUs) face the highest risk of road accidents. Various innovations aim to enhance their safety, including the development of Automated Vehicles (AVs). As AVs get integrated into traffic, VRUs might interact with them at crossings. To ensure efficient development of AV, it is crucial to understand how VRUs adapt their behaviour and make road-crossing decisions infront of AVs. This paper is part of a longitudinal Virtual Reality (VR) experiment where participants crossed a road under varying environmental conditions. Interviews were carried out using self-confrontation technique, where participants watched a video of their road-crossing, then reflected on their decisions. This method helps explain quantitative data but rarely used in pedestrian-AV studies. Inductive content analysis was used to analyse the transcripts, categorising findings into two dimensions. The findings reveal that participants’ decision-making varied across the scenarios and experimental groups. The two general dimensions- human-computer interaction and user experience can guide the design of a framework to understand pedestrian-AV interactions. The analysis provided deeper understanding of pedestrians’ crossing behaviour, highlighting factors influencing decisions and behavioural adaptation. Gaps in AV communication were identified and design recommendations for signalling, emphasising the need for intuitive, pedestrian-friendly AV behaviour to enhance safety was provided.